Abstract. I perform a quantitative comparison of the shape of the optical luminosity functions as a function of galaxy class and filter, which have been obtained from redshift surveys with an effective depth ranging from z 0.01 to z 0.6. This analysis is based on the M * and α Schechter parameters which are systematically measured for all galaxy redshift surveys. I provide complete tables of all the existing measurements, which I have converted into the UBVR c I c Johnson-Cousins system wherever necessary.By using as reference the intrinsic luminosity functions per morphological type, I establish that the variations in the luminosity functions from survey to survey and among the galaxy classes are closely related to the criteria for galaxy classification used in the surveys, as these determine the amount of mixing of the known morphological types within a given class. When using a spectral classification, the effect can be acute in the case of inaccurate spectrophotometric calibrations: the luminosity functions are then biased by type contamination and display a smooth variation from type to type which might be poorly related to the intrinsic luminosity functions per morphological type. In the case of surveys using multi-fiber spectroscopy, galaxy classification based on rest-frame colors might provide better estimates of the intrinsic luminosity functions.It is noticeable that all the existing redshift surveys fail to measure the Gaussian luminosity function for Spiral galaxies, presumably due to contamination by dwarf galaxies. Most existing redshift surveys based on visual morphological classification also appear to have their Elliptical/Lenticular luminosity functions contaminated by dwarf galaxies. In contrast, the analyses using a reliable spectral classification based on multi-slit spectroscopy or medium-filter spectrophotometry, and combined with accurate CCD photometry succeed in measuring the Gaussian luminosity function for E/S0 galaxies. The present analysis therefore calls for a more coherent approach in separating the relevant giant and dwarf galaxy types, a necessary step towards measuring reliable intrinsic luminosity functions.